TY - JOUR TI - Automated bridge deck evaluation using ground penetrating radar scans DO - https://doi.org/doi:10.7282/T37M05Z3 PY - 2013 AB - Reinforcement concrete (RC) bridge decks are surveyed regularly to ensure that they are safe to use and to determine if they require rehabilitation or replacement. The bridge surveys include evaluating subsurface bridge condition. RC bridges have steel reinforcement bars, also called rebars, embedded in their surface, which are prone to corrosion due to factors like moisture, carbonation, use of deicing salts and aging. By the time the effect of corroded rebars is visible on deck surface in form of cracks, the damage is tremendous. If left unchecked, corroded rebars can deteriorate at a faster and significantly affect bridge integrity. So, it is very important to timely identify and repair deteriorated rebars. Ground Penetrating Radar (GPR) is a widely used non-destructive technology (NDT) for detecting subsurface anomalies in variety of structures including RC bridges. The raw GPR data is represented as images that can be processed for obtaining a deterioration map of a bridge, which indicates the level of corrosion in rebars for the entire bridge. The existing methods to generate the deterioration map using GPR data are semi-automated, time consuming and depends on expertise of the engineer analyzing the data. In this thesis, we work towards automating the process of obtaining deterioration map of RC bridge decks based on measuring signal attenuation at the upper rebar mat using GPR. Intensity and gradient-based feature vectors were explored to construct a classifier, which can detect the regions of interest (ROI) corresponding to each rebar in images. Each classifier was tested on datasets constructed from two different bridges. Further, the exact location of rebar was found in each ROI. Once all the rebars were detected throughout the bridge, depth-correction of the measured attenuation is applied so that the component of that measured attenuation caused solely by variation in rebar depth does not skew the results. Finally, a deterioration map was generated which indicates the level of corrosion in the bridge. The proposed algorithm was tested on two RC bridges and the deteriorated regions obtained are compared with the results obtained using existing tools. KW - Electrical and Computer Engineering KW - Concrete bridges--Maintenance and repair KW - Concrete bridges--Inspection KW - Ground penetrating radar KW - Reinforced concrete LA - eng ER -